Login
You're viewing the functional.cafe public feed.
  • Nov 10, 2026, 7:05 PM

    (b) this is totally anecdotal, but as someone who has been teaching computer programming to adults for ~20yrs or so, the biggest obstacle that students face is not that the material is difficult (even when it *is* difficult), but that students are *culturally predisposed to a belief that they're not smart enough to program*, which acts as a permission structure for giving up. (this is peculiar, because no one having trouble in a knitting class thinks "i guess i'm not smart enough to knit")

    💬 24🔄 79⭐ 0

Replies

  • Nov 10, 2026, 7:09 PM

    (c) programmers are predisposed to convincing others that programming is too complicated for "normal people" as a means of job security and achieving social status; computer manufacturers and their software-making friends are predisposed to a marketing message that states: computers are too difficult for normal people, so buy this new machine/software, which is finally the computer/software that normal people can use! together both sides form a kind of mutual ratchet imo

    💬 19🔄 29⭐ 0
  • Nov 10, 2026, 7:11 PM

    (the trope of the "computer nerd" in media is such an effective embodiment of both sides of this rhetoric that its prevalence almost seems like a conspiracy imo)

    💬 4🔄 8⭐ 1
  • Nov 10, 2026, 7:16 PM

    anyway, whenever someone says something like "LLMs have finally made programming accessible to normal people!" I try to remember that the Silicon And Fossil Fuel Pyramid Scheme Industry has spent literally hundreds billions of dollars in an effort to convince you that computers were *already difficult to use* (a grift that computer programmers are largely in on!) and to politicize LLMs as being "easy to use" in contrast

    💬 20🔄 45⭐ 1
  • Nov 10, 2026, 7:34 PM

    @aparrish Presented to you by the same people that convinced (almost) everyone that simple text and images need proprietary container formats

    💬 0🔄 0⭐ 24
  • 💬 2🔄 1⭐ 18
  • Nov 10, 2026, 10:37 PM

    @sinza @aparrish But there should be some meritocracy right? Given two patches, if one is objectively better it should be chosen, right?

    The other week I continued trying to help someone learning to code (sadly she's ai-pilled and not getting anywhere). She said her friend said I was so intelligent. Later, to explain the bell curve using IQ as an example, pointed out what it means. She would not accept people could be at the bottom despite suggesting I was towards the top. How to explain?

    💬 5🔄 0⭐ 1
  • Nov 10, 2026, 10:52 PM

    @bbuk I'm the wrong person to ask, as I actually oppose the Open Source movement and find it detrimental to software freedom. I also don't like the FSF even though I do like software freedom.

    That being said, assuming moral equivalency, the better patch would be better, but the idea of good code is more subjective than one would think.

    As to explain, I wish I knew because I struggle with that myself thanks to years of being told I'm smarter than others despite logically knowing that intelligence is way more multifaceted than what IQ says.

    @aparrish

    💬 0🔄 0⭐ 4
  • 💬 0🔄 1⭐ 9
  • Nov 11, 2026, 12:17 PM

    @bbuk @sinza @aparrish No. Given two patches, if one is from someone who will destroy your community and the other has errors, you ignore the former and work with the author of the latter to get it right. This is not hard.

    "Meritocracy" is complete and utter bullshit.

    💬 1🔄 1⭐ 0
  • Nov 11, 2026, 3:29 PM

    @dalias @bbuk @sinza @aparrish the problem with just saying one thing of two patches is "better" is that the statement obscures a lot of factors. Better for whom? For the maintainer? For the community? Better in the longer term or shorter term? The judgement of "better" is a matter of values of both the individual and the community they belong to.

    💬 0🔄 2⭐ 0
  • Nov 11, 2026, 1:38 PM

    @bbuk @sinza @aparrish You can write your own code without contributing it upstream. Ideally more people should write private code that suites their own workflows and private needs.

    If a workflow becomes refined enough to be worth sharing then you do. Then the community gives you polite suggestions on how the code can be further polished. This is basically how the Emacs community functions. There may be multiple similar solutions and thats ok, not everything needs to gets merged upstream

    💬 0🔄 3⭐ 1
  • Nov 11, 2026, 5:01 PM

    @bbuk @sinza @aparrish it's weird that you use the word meritocracy but don't even consider what the word merit means. That should be your starting point, not some silly hypothetical using the phrase "objectively better." objectively how? what subjective process has determined that this one is "objectively" better?

    💬 0🔄 1⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Nov 10, 2026, 7:45 PM

    @aparrish I think we could do a lot to:

    1) Make programming feel more accessible and less hostile
    2) Make it actually more accessible (through simpler languages, better documentation, higher-level abstractions, nicer interfaces)
    3) Educate people to increase their skill

    Simultaneously lower the floor, extend a ladder, and encourage people to increase their confidence.

    💬 4🔄 14⭐ 50
  • Nov 10, 2026, 7:48 PM

    @aparrish As far as job security goes, more people should learn to program and more programmers can learn to do more complex things.

    Programmers can move from being expected to run and maintain the car to designing and innovating new techniques and paradigms.

    This also applies to LLMs since they can only (badly) maintain the car but don't have the capacity for this level of expertise. They can glue but can't engineer.

    💬 0🔄 4⭐ 16
  • Nov 10, 2026, 8:00 PM
    @liberty @aparrish I agree with all of the above. I've been programming since the '80s, and another problem I've seen is that over time the hardware itself has made programming more difficult, locking things down and actively hiding the internal workings. Take the dumpster fire that is UEFI, for instance.
    💬 1🔄 0⭐ 0
  • 💬 2🔄 0⭐ 0
  • Nov 10, 2026, 10:31 PM

    @shriramk Yes.

    Programming—at least conceptually—is not as difficult as many make it out to be.

    At the same time, the tools have been made less accessible by the same people who want to perpetuate the myth that it's not something that the average person can/should do.

    💬 2🔄 0⭐ 0
  • Nov 10, 2026, 11:42 PM

    @me @shriramk Consider also: we don't have to fight with memory management anymore, but we do have to fight with npm and whatever the latest shiny thing is. The two classes of problem are very different

    💬 2🔄 0⭐ 0
  • Nov 11, 2026, 3:11 AM

    @jmopp @me I'm well aware — I've been programming with GC since 1989 and been building languages with GC since 1995. But even I find the current package management world (to say nothing of the security issues they introduce) pure hell. On the one hand they enable people to build what they want, on the other hand *do* overwhelm people into a kind of numb submission and acceptance (leading to exploits).

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Nov 11, 2026, 3:09 AM

    @me Programming is harder than most CS1 people make it out to be, especially as traditionally done. State is hard: it introduces time; aliasing is hard: it introduces space. The combination is such that even experienced programmers mess up, which is why we *need* languages like Rust in the first place.

    I've spent 4 decades trying to spread programming widely, even to middle-schoolers, with great success — so I'm not a gatekeeper. Of course avg people can do it. But doesn't make it easy.

    💬 0🔄 3⭐ 0
  • Nov 11, 2026, 4:15 PM

    @shriramk @me It's not that it's harder, it's that there's more crap to deal with. It's the difference between difficulty and obfuscation.

    It's like picking up a note you wrote last week that's sitting on a clean desk, versus picking up a note you wrote last week from a desk covered with piles of paper.

    💬 1🔄 0⭐ 0
  • Nov 11, 2026, 8:53 PM

    @mathew @me "More to deal with" absolutely makes things harder. We have limited cognitive capabilities, and they get overwhelmed.

    Concurrency, for example, forces us to deal with all possible interleavings (in the worst case). That's not "crap", that's just the nature of the thing. (I go to great lengths to parallel programming because I have that luxury, but a lot of people don't, and some who do can't just use SIMD, say.) Or async. They're there for good reasons, not crap. But they're hard.

    💬 1🔄 0⭐ 0
  • Nov 12, 2026, 4:22 PM

    @shriramk @me We're talking about beginners, though. They don't need to leap into concurrency. They just need to be able to put together some simple UI and draw stuff on the screen.

    That said, the fact that so many languages make concurrency a dangerous chore is unfortunate. If functional programming languages were more accessible and the nursery pattern was more widely supported…

    💬 1🔄 0⭐ 0
  • Nov 12, 2026, 9:17 PM

    @mathew @me Kids can make a cat walk back and forth across a screen in Scratch in an hour. And then what? Presumably the programming we're talking about here is actually *doing* something, perhaps of value that solves some other problem they have. Otherwise we can define programming down to almost nothing.

    Why do you think FP isn't accessible? I co-run one of the US's largest in-school computing outreach programs, and we do nothing but pure FP! So maybe we have different meanings of "FP".

    💬 1🔄 0⭐ 0
  • Nov 13, 2026, 12:22 AM

    @shriramk @me If you’re going to say that they have to be able to solve some real problem easily and also that they should use an FP language with solid concurrency support, I’m going to have to be very skeptical.

    💬 1🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Nov 10, 2026, 8:02 PM

    @liberty

    The problem with simpler languages is that you have to sacrifice a lot in order to make a language simple: run-time performance, safety, and so on.

    Rust is by most accounts a very non-simple language, but I wouldn't part with it. Why? Because it makes my job more secure? No; my job isn't in any danger. It's because Rust very often stops me from making mistakes, and it does so without making my program slow and waste memory.

    @aparrish

    💬 5🔄 0⭐ 11
  • Nov 10, 2026, 8:12 PM

    @argv_minus_one @aparrish @liberty Any time someone tells me that one language—any language—is the best, I get suspicious.

    Admittedly, I don't know a lot about Rust, though I have dabbled in it. It's my experience that the best programmers know multiple languages and can pick the one that best suits the task at hand.

    That said, I also strongly believe that the "best" language is the one that works for you, so in your case... that'd probably be Rust. 🙃

    💬 1🔄 0⭐ 0
  • Nov 10, 2026, 9:46 PM

    @me

    Yeah, I'm always anxious about bugs, and Rust does a ton of correctness checking, so I tend to use it when feasible.

    Although I am worried about its long-term viability, seeing as its LLM policy is not a complete ban. It'll lose its reliability real quick if they start vibe-coding the compiler…

    @aparrish @liberty

    💬 0🔄 0⭐ 7
  • Nov 10, 2026, 9:49 PM
    @argv_minus_one Yeah, I use Haskell for similar reasons (the type system saves me from a lot of potential bugs) though I've looked at rust for scenarios where system resources are scarce. I actually wish I understood it better.
    💬 2🔄 0⭐ 0
  • Nov 11, 2026, 3:39 AM

    @me I read somewhere that there's also LLM output in GHC these days. (one of the reasons I want to eventually make my Spiteful Lisp C Compiler share a backend with a Spiteful Lisp Haskell Compiler and a Spiteful Lisp Rust Compiler - the other reason being that I tried and failed to port GHC to Haiku)

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Nov 10, 2026, 8:12 PM

    @argv_minus_one @aparrish We'll always have higher and lower level languages. We can get rid of some of the cruft of the more complex languages, but people don't have to be expected to use them or deal with complexities like memory management.

    Some concepts are fundamentally harder and will require more expertise and domain-specific tools. But that's not all programming.

    I agree that Rust is pretty nice for the more complex situations (but more complicated than it needs to be still IMO).

    I am fine sacrificing some run-time performance in direct control over the computer for accessibility. I am a fan of declarative programming and functional programming as avenues for this.

    💬 1🔄 0⭐ 7
  • Nov 11, 2026, 11:35 AM

    @liberty
    @argv_minus_one @aparrish
    AFAIK the most popular programming language is Microsoft Excel. A lot of non-programmers use it for stuff that I would do with a shell one-liner or a one-off python script.
    I think there's room for improvement here.

    💬 0🔄 1⭐ 0
  • Nov 10, 2026, 10:09 PM

    @argv_minus_one @liberty @aparrish
    "Simpler" languages almost always just shift the complexity onto the programmer. So they achieve a superficial sense of simplicity, with the complexity distributed all over codebases.

    Rust is a great example. There's a fair bit of complexity in the types, but the result is that everyone who can follow its discipline benefits from overall clarity and simplicity. Similar programs in other languages distribute thinking about aliasing and ownership.

    💬 0🔄 0⭐ 6
  • Nov 10, 2026, 11:32 PM

    @argv_minus_one @liberty @aparrish

    Writing safe accessible languages is hard, but we should keep trying.

    I once taught a kid at my local library how to code using Scratch.

    Within about 20 minutes, he re-invented fork-bombing. (Future Cybersecurity engineer, I imagine.)

    The Scratch engine handled his fork bomb perfectly - filling his screen with the expected randomized bat animations, while quietly cleaning up extras when there were too many for thre processor to handle.

    Building good accessible languages is certainly a challenge, but I believe it is achievable. And it's so important to help us teach fork bombing to our next generation of Cybersecurity engineers.

    💬 1🔄 0⭐ 2
  • Nov 11, 2026, 4:42 AM

    @EdTheDev

    Writing a safe accessible language seems pretty impossible, though, because to do so would be to reconcile two mutually-opposing forces.

    Safety requires a sophisticated type system, with generics and bounds and all that. Rust has that, as do some other languages, but such languages are famously non-simple.

    A dynamically-typed language is simpler, but then you lose type safety, and have to keep track of types in your head, which is easy to screw up.

    @liberty @aparrish

    💬 1🔄 0⭐ 2
  • 💬 1🔄 0⭐ 2
  • 💬 0🔄 0⭐ 0
  • Nov 11, 2026, 5:24 AM

    @argv_minus_one @liberty @aparrish

    As I mentioned, the solution is animated bats.

    (Sorry, I'm not treally trying to argue, just trying to get a laugh.)

    I find that new programmers that I teach need accessible languages, rather than safe ones, or even fast or powerful ones.

    They need languages with great tutorials, interesting uses, and high quality error messages.

    💬 0🔄 0⭐ 3
  • Nov 10, 2026, 11:42 PM

    @argv_minus_one @liberty @aparrish

    Scheme is a very simple language, safe, and has native compilers that make it fast enough for most of what people actually need (If you need scientific number crunching Julia is a secret LISP that does that full speed)

    💬 1🔄 0⭐ 2
  • Nov 11, 2026, 2:27 AM

    @dlakelan

    I'm not familiar, but Wikipedia says Scheme is dynamically typed, which is not great for type safety.

    Full-featured static type systems are complicated, what with generics and bounds and all, but once you learn them, you can rely on them and so can everyone else who works with your code. Dynamic typing increases cognitive load and bug probability *a lot*.

    @liberty @aparrish

    💬 1🔄 0⭐ 1
  • Nov 11, 2026, 4:53 AM

    @argv_minus_one
    I have not found that to be true at all. Ive worked in statically typed languages, Pascal, C, C++, a tiny bit of ML and in dynamically typed languages, Scheme, R, Python, Prolog, CommonLisp, Erlang, Julia

    In general bottom up programming in REPL based languages has been far less bug prone for me. The kinds of bugs type systems prevent are rarely the ones I am concerned about. its more like doing wrong behavior that worries me. Seems like a preference mostly.
    @liberty @aparrish

    💬 0🔄 0⭐ 2
  • Nov 11, 2026, 2:49 AM

    @liberty @aparrish in my anecdotal evidence any approach to simplification is met with resistance and exclusion in the form of 'simple alternative is not for solving real issues'

    Think of the reputation of PHP or the impact that abstractions like snap had or didn't have.

    Also it feels like while we've made building more and more complex architectures possible over time we failed to build abstractions that make simple things easy to do.

    💬 1🔄 0⭐ 2
  • Nov 11, 2026, 2:59 AM

    @mainec @aparrish That's nearly the universal response and they have a point. If you abstract things you nearly universally have to get rid of context underneath.

    I think we solve this by both cleaning up the excess complexity and by being careful about what we're abstracting.

    We shouldn't make any level of abstraction have a higher barrier to entry than necessary.

    And those abstractions also need to be carefully considered with both technical skill and care for the end user. That's what PHP lacked.

    💬 0🔄 0⭐ 1
  • Nov 10, 2026, 7:49 PM

    @aparrish I mostly agree with you here, especially on the "LLMs make computers accessible" tripe, but I do think there's something about the *way* computers are user hostile and complicated that is different. I say this mostly from many years working with non-tech people, and especially my parents. There is *something* different going on there. It's not nearly as large or insurmountable as CompSci people, mass media, or the AI bros want to pretend, but there's something.

    💬 2🔄 0⭐ 7
  • Nov 10, 2026, 7:51 PM

    @aparrish I *think* it's about the "virtualness" that makes it feel more like a black box. Yes, absolutely, moderate skill with the command line is no more complicated than moderate skill in the kitchen; I could teach someone to be a programmer more easily than a farmer. But there's something about the ways in which it fails, what you can see and get your hands on, that makes it more intimidating to many people; I think the broader phenomenon is based on taking that and exacerbating it.

    💬 2🔄 2⭐ 11
  • Nov 10, 2026, 8:08 PM

    @a @aparrish I'm not a coder but am a fairly skilled user. Most 'major' software these days seems designed to be actively hostile to users being able to use and adapt it as we want. Wasn't always that way but it is now. So I suspect that experience with software which actively fights against what you want to do may be part of what leaves people feeling they aren't capable.

    💬 0🔄 0⭐ 12
  • Nov 11, 2026, 12:18 AM

    @a @aparrish yeah, i think that is a huge part of it. software being on a screen instead of a physical object in the real world can make it (relatively speaking) really difficult to observe the state of the system and discover its capabilities! and there's a big hurdle to overcome in learning user interface patterns, ways to figure out how to do things you don't know already, how to troubleshoot when things don't work, etc

    💬 1🔄 0⭐ 0
  • Nov 11, 2026, 12:44 AM

    @a@pdx.social @aparrish@friend.camp @b@mrrp.place

    The fact that computer science/programming is artifice all the way down is often overlooked. Overlooking that purposefully to perpetuate the narrative that “this is the way things are” is both useful for the aforementioned people and harmful for accessibility.

    The human-designed structures involved, from hardware to operating systems to libraries to applications, were designed in a certain way for a reason. Approaching them from a completely abstract angle is very confusing, but once you know the history around them, they become much more understandable. The fact that these were all designed by people also means they can be redesigned by people for approachability, which is something a certain kind of crusty old programmer, SaaS business, or CS academic hates to hear.

    💬 0🔄 2⭐ 0
  • Nov 11, 2026, 9:53 AM

    @a @aparrish

    My take (from working with people who don't know computers very well) is that there's a few barriers to programming in addition to the cultural ones (which I am inclined to agree are very significant).

    • A lot of applications -- before you even get into programming -- are really badly designed or even downright user-hostile. Think a bazillion configuration settings to turn off tracking and the like. Then when you start programming it's treated as extra-difficult.
    • There's programming as in "writing a program in C to calculate something" and "getting a fully featured program that does the things you want to do", and those are not the same thing, I've seen people trip on the gap between basic programming and more complex stuff. There is also surprisingly few resources for bridging that gap.
    • Complexity can show up where you don't expect it. For example, I wanted to know "how do I write a grammar extension in LibreOffice" and expected there to be a walkthrough on setting that sort of thing up, but instead only found a bunch of Java-Object-Ese with no explanation of why it's set up that way
    💬 0🔄 0⭐ 0
  • Nov 10, 2026, 7:53 PM

    @aparrish probably worth also considering the gendered nature of the computer nerd and how the switch from programming as "women's clerical work" to something more like "Smart Science Man Stuff" came about as personal computers were hitting the market. as in maybe perpetuating the idea helps shore up masculinity, so it is in many people's interests to do so even if they're not computer nerds themselves?

    💬 1🔄 1⭐ 0
  • Nov 10, 2026, 9:49 PM

    As somebody working in that field I have to say I wish we had a more equal gender distribution. I am not sure exactly when it went wrong, by the time i entered the field it was already very skewed.

    💬 2🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0